A COMPUTATIONAL MODEL OF AUDITOR KNOWLEDGE AND REASONING PROCESSES IN THE GOING-CONCERN JUDGMENT

This paper presents a computational model of auditor expertise within the domain of the going-concern judgment. The model proposes that auditors possess and use three broad categories of knowledge: financial knowledge, event knowledge and procedural knowledge. The model, referred to as GCX, has been implemented in a computer program that processes five years of data from a real-world company and makes going-concern judgments similar to practicing auditors. There are three primary contributions of this research. First, the model proposes an overall structure of auditor financial and event knowledge in the going-concern task. In particular, GCX proposes that knowledge of financial measures is causally linked to knowledge of actual events via abstract models of company finances and operations. In addition, the model proposes that some actual events are linked to more generalized knowledge of what normally happens when these events occur. Such a memory structure implies that an important component of auditor expertise in the going-concern judgment involves knowledge of specific events related to particular clients and the ability to reason about those events. Second, GCX proposes four reasoning processes that operate on financial and event knowledge to perform the going-concern task: problem recognition, causal reasoning about problems, evaluative reasoning about plans to mitigate problems, and a process that renders one of three going-concern judgments. The model is based on a complex indexing of financial and event knowledge structures that supports the four reasoning processes. Third, and perhaps most important, the GCX model proposes that a central component of the going-concern judgment is causal reasoning based on knowledge of actual client-related events. Interviews with auditors demonstrated that they had extensive knowledge of their client's operations, their client's industry, and world events that affected their client's financial problems. This causal reasoning based on event knowledge played a critical role in determining whether a financial problem was significant and whether management's plans would mitigate the problem. Thus, it is proposed that client-based knowledge plays a critical role in auditor judgment and that research into auditor expertise that ignores this type of knowledge is incomplete. The paper concludes by inferring from the model a number of testable predictions and research questions for further research.